In recent days, a strange new ChatGPT trend has been sweeping TikTok. Right-wing Christians perform an ecstatic devotion they call speaking in tongues, and then try to get ChatGPT to guess the language they used. Some claim ChatGPT has told them they spoke in exotic Ancient Near Eastern and Mesopotamian languages. But they’re wrong. Today, let’s examine this trend, see why ChatGPT can’t actually validate their babbling, and see why these Christians need this validation so much.

(This post first went live on Patreon on 7/21/2025. No audio recording tonight, my throat is still delicate. But I’m working on some new changes to my setup that might be very pleasing indeed—stay tuned!)

SITUATION REPORT: Speaking in tongues meets the AI age

A recent article from Christian Post caught my eye last week: “Can ChatGPT interpret speaking in tongues?” Written by their Senior Reporter Leonardo Blair, it covers an unusual development in right-wing Christianity: Shawn Bolz, a self-styled prophet associated with Bethel Church in Redding, proclaimed in a video that ChatGPT can apparently translate speaking in tongues.

In a video dated June 29, Shawn Bolz described the trend:

ChatGPT—several people are doing this. It’s become a YouTube trend where people are speaking to in their spiritual language that they have from God, and it’s actually interpreting some of the words for them, sometimes whole phrases for them. One of the people I saw online it was ancient Sumerian that somehow ChatGPT understood!

We’re seeing AI catch up to some of the spiritual fruit that we have as Christians, especially those of you who are charismatic or Pentecostal and you guys believe in speaking in tongues. It’s pretty profound. Well, AI has been writing and building and translating and diagnosing all kinds of things. [Source: YouTube, 2:45 timestamped]

This view contrasts dramatically with some right-wing Christians’ fears of demons animating AI chatbots.

Though many Christians seem very doubtful of his claims, others applauded this trend. As one commenter to Bolz’s video wrote:

From AI to alternative currencies, Shawn opens our eyes to how God wants to partner with His people in innovation—not just to survive change, but to lead it. He challenges us to stop playing catch-up and start co-laboring with God in the systems that are shaping our future.

But this unusual trend turns out to be yet another false lead for prophets and overenthusiastic believers alike. Not only can ChatGPT not do what they’re imagining, but speaking in tongues itself isn’t what they believe it is.

Speaking in tongues: A quick introduction

Many right-wing Christians, called charismatic Christians or just charismatics, practice speaking in tongues. It is technically called glossolalia. Charismatics claim the practice comes an event described in Acts 2, in which the Holy Spirit enabled many Jews to start miraculously speaking in other languages that visiting foreigners recognized. Today, it mostly serves as an ecstatic form of prayer.

Despite claims, speaking in tongues has no verified connection to any real languages, past or present. Most believers mimic sounds from what they imagine are 1st-century Ancient Near Eastern (ANE) languages. In the following video, a guy coaches a new convert in how to do it:

Usually, speaking in tongues looks way more energetic than that, though. For a more representative look at the practice, here’s Paula White speaking in tongues (satirized by Betty Bowers as nonsense):

Again, none of what these Christians are doing has ever been verified to be any real language. Back in the 1980s and 1990s, when I was Pentecostal myself, we sure thought that it was. Rumors abounded of this or that Pentecostal whose tongues language was a real live language. My own church had a very young girl in the congregation whose prayer language was apparently Aramaic (a Semitic language thought to be Jesus’ mother tongue). As the legend went, an unnamed Jewish scholar once visited my church, heard her praying, and reported to our pastor that he recognized what she was saying.

When I first heard this striking claim, I immediately suggested to my pastor that we invite some linguists in to record her and verify this miracle. My pastor had quite a tough time dissuading me. It took years for me to understand why he might not have been wild about testing her tongues language.

After fundamentalism’s fusion with evangelicalism, which completed by the late 1990s, the newly-fused “fundagelicals” seemed content to consider their practice a private divine language. Instead of being testable and falsifiable, now speaking in tongues could be entirely between them and Jesus. It no longer needed to be a real language spoken at some point in human history. But many charismatics clearly want it to be real.

(Speaking in tongues plays very different roles besides forming part of a charismatic’s prayer devotions, but for our purposes today I’m discussing it only in that context.)

Christians claim ChatGPT agrees with them about tongues being a real language

In his video, Shawn Bolz refers to a May video posted by Renée M. Simpson, a life coach going by Well Diva on TikTok. Her video is now deleted, but numerous content creators discussed it online (like “THE PREACHER’S PORTAL” on YouTube). In the video, Simpson tearfully reveals that ChatGPT had accurately “translated” her babbling. She claims that the AI chatbot knew what her babbling meant. Notably, she doesn’t specifically say what exact language ChatGPT had assigned to her babbling. She only says:

It [the supposed translation] was very, very specific. It was definitely interpreting what was on my heart in prayer, and nobody could have known that. [Source: YouTube, timestamped 1:05]

Though obviously many Christians reacted with alarm to Simpson’s claim, others wholeheartedly embraced it.

One failed ChatGPT experiment

Here’s “SkyeProphet” trying out the trend.

Even though he slowed down his rapid-fire ANE-ish tongues-talking, ChatGPT correctly assessed what it had received as glossolalia. SkyeProphet then demanded that it “comment on any linguistic similarities from the excerpt in relation to known human languages.”

ChatGPT failed to play along. So he told it to forget that the audio excerpt he provided was glossolalia and assess it instead as a real language. In response, it told him his recording contained common elements found in many languages, including English. It correctly informs him that what he provided lacked any semblance of “clear grammatical structure.”

After even more questioning, ChatGPT conceded that some of the excerpt’s “phonetic patterns” might sound somewhat similar to “Semitic languages.”

That’s not surprising. When I was Pentecostal, I heard the exact same “phonetic characteristics” many thousands of times. They’re all trying to sound like they’re speaking Aramaic. I’ve never heard of any charismatic doing anything that sounds like, say, Occitan, Tamil, Gaelic, or Attic Greek.

Eventually, ChatGPT offered a few halfhearted-looking assessments of the supposedly Semitic phonetics in the excerpt. This led SkyeProphet to conclude that “if you were just hearing a message in tongues and you understood Semitic languages, you might be able to pick up some of that [speaking in tongues].” He completely misinterpreted ChatGPT’s responses to fit his own narrative.

As miracles go, this one falls far short of divine. But then, SkyeProphet claims in a comment that he doesn’t believe tongues-talking is a real human language, just that it’s a language.

In that case, it’s odd that no charismatics have ever mimicked animal communication—like, say, whalesong.

That’d be an impressive bunch of noises in any Pentecostal church!

Like all trends, ChatGPT tongues-translation died out pretty quickly

It took a few weeks for Shawn Bolz to pick up on this trend. By then, it seemed like most glossolalia believers had already weighed in on one side or the other. As I said, most condemned the idea of using ChatGPT to translate tongues.

Some condemned it because of their fears of demons. Simpson herself addressed that concern in a follow-up video on TikTok.

Other Christians, like Lethatron James, correctly explained that ChatGPT is just a predictive large language model (LLM). It learned what it knows from studying actual languages as used by humans. It uses that knowledge to predict patterns in language—which it then uses to respond to users. If speaking in tongues doesn’t conform to that model, ChatGPT won’t be able to translate it. It failed to translate his own tongues-talking.

Similarly, Leonardo Blair, who wrote the Christian Post story that started our topic today, had the same things to say. At the end of his article, he notes that he himself is Pentecostal. ChatGPT failed to translate his tongues-talking either, which he seems to have expected. He knows it’s just a chatbot—a sophisticated one, yes, but still a chatbot.

What this ChatGPT trend means in greater context

I also liked that James alludes to what’s really going on here: an appeal to authority. The appeal to authority is a logical fallacy. It seeks to establish a claim’s veracity through an authority figure’s agreement with it. It’s a substitute for actually supporting a claim.

So often online, I see people talking about going to ChatGPT or other similar AIs with questions, then offering the results to others as a definitive answer. It happens so often that it’s recently become a meme of its own. In the meme, people say they asked some other source like “Alucard of Wallachia, son of Dracula” or their orange cat, then relay the supposed answers these sources provided.

"i asked chatgpt" "i asked grok" I asked alucard of wallachia, son of dracula and he said [screenshot of alucard anime subtitled 'grunts in confusion']
"i asked grok" "i asked chatgpt" well i asked yun'shul and they said "P̶L̶U̷C̸K̸ ̴O̷U̵T̶ ̸T̷H̵I̵N̶E̵ ̷E̵Y̶E̷S̴ ̸T̷E̵A̵R̸ ̸O̶U̶T̵ ̶T̶H̸I̸N̵E̷ ̵T̶H̸R̸O̴A̵T̷ ̸L̶A̵Y̵ ̶W̸A̷S̵T̵E̷ ̸T̵O̶ ̶A̷L̷L̴ ̵W̴H̵I̴C̶H̴ ̷I̸S̷ ̵S̷A̵C̶R̵E̶D̷ ̶T̵H̵E̴ ̵M̴O̵O̶N̵S̸E̶Y̷E̶ ̷C̵O̴N̷V̷E̴R̶" [screenshot of weird anime or horror-game's shrine]
“i asked chatgpt” “i asked grok” i asked cheese n he said meow meow meow meow meow :D [two pictures of a curious and uncertain orange cat]

This meme pokes a little fun at our very human desire for validation from some higher authority. It’s not bad to want that validation. Nor is it always bad to accept claims based on authority. Many claims accepted in this way save us valuable time without harming us.

To some extent, all of us do it—as I learned years after deconversion.

How my office completely surprised me with a big award

A long time ago, Mr. Captain and I worked at the same call center and in the same management team. My company audited many of our calls for performance—with perfect scores being extremely rare. Any time someone got a perfect score, the entire building celebrated with a surprise procession of upper management, baked goods or candy, and lots of cheering.

I’d earned a rare perfect score. But our building’s head managers, away for an extended time, were afraid I wouldn’t be surprised by the time they got back for the procession.

Mr. Captain, who wanted me to have this honor, said “Leave it to me. I’ll make sure she’s surprised.”

He and our supervisor tricked me into thinking our company’s employee portal—which contained all our stats, including audit scores—was out of commission. For weeks, I overheard the two of them complaining about the supposedly-broken portal. I trusted their word and didn’t like wasting time by pulling it up anyway, so I didn’t check it for myself. At the time, I was just happy nobody could possibly get peevish at me for not checking it!

When the head managers got back, their procession completely surprised me. But along with that surprise, I’d also gained valuable insight in how easy it is to accept appeals to authority—especially when we have no reason to doubt them.

Testing the appeal to authority

In cases where the information’s accuracy is more important, my reaction differs enormously. I didn’t believe Princess Diana had really died until I heard it from reputable news sources. Similarly, I quickly broke free of AIDS denialism thanks to vetting my sources more carefully. Learning about neuroscience from unbiased sources helped me shed my lingering afterlife beliefs.

All too often, Christians fall prey to bias, faux-expertise, and self-interest. Their leaders indoctrinate them to accept appeals to authority as valid supports for claims. If someone they consider an authority figure accepts an idea, that makes it extra-true. So Christians who think speaking in tongues is a real actual language will be primed to accept what their thought leaders say about it. Just as Pentecostals didn’t care what Southern Baptists thought about speaking in tongues in the pre-fusion years, they won’t care about outsiders criticizing their practices.

As AI continues to increase in use and acceptance, Christians will continue to bend it to their needs. But real knowledge and wisdom require more than appeals to authority. And this trend tells us they don’t have what it takes.

Please support my work!

Thanks for reading, and thanks for being part of our community! Here are some ways you can support my work:

  • Patreon, of course, for as little as $2 a month! I now write Patreon posts twice a week. They drop on Tuesday and Friday mornings for patrons, then a few days later on the main site, Roll to Disbelieve.
  • Paypal, for direct one-time gifts. To do this, go to paypal.com, then go to the personal tab and say you want to send money, then enter captain_cassidy@yahoo.com (that’s an underscore between the words) as the recipient. It won’t show me your personal information, only whatever email you input.
  • My Amazon affiliate link, for folks who shop at Amazon. Just follow the link, then do your shopping as normal within that same browser window. This link adds nothing to your Amazon bill, but it does send me a little commission for whatever you spend there.

And as always, sharing the links to my work and talking about it!

Visited 1 times, 1 visit(s) today

Captain Cassidy

Captain Cassidy is a Gen-X ex-Christian and writer. She writes about how people engage with science, religion, art, and each other. She lives in Idaho with her husband, Mr. Captain, and their squawky orange tabby cat, Princess Bother Pretty Toes. And at any given time, she is running out of bookcase space.

1 Comment

Evangelical thrilled because ChatGPT said his god is totes for realsies - Roll to Disbelieve · 08/18/2025 at 4:01 AM

[…] really is the all-flavor end run for evangelicals these days. One of their more recent trends was trying to get this Large Language Model (LLM) AI to “interpret” them speaking in […]

Leave a Reply

Avatar placeholder

Your email address will not be published. Required fields are marked *